ESMO 2016: PD-L1 Expression as a Biomarker for Checkpoint Inhibitors Response in Patients with Advanced Melanoma

A pooled analysis from 3 clinical trials in advanced melanoma shows that while nivolumab and nivolumab plus ipilimumab patients with ≥5% PD-L1 tumour expression have similar progression-free survival (PFS), durable response rates were higher for the combination of both agents versus nivolumab alone, across all PD-L1 expression subgroups. Given that treatment-related side effects with the nivolumab plus ipilimumab combination are higher than seen with nivolumab when used alone, there has been interest in whether PD-L1 testing could be used to better understand the benefit risk of both treatment options said Georgina Long, from the Melanoma Institute Australia, The University of Sydney, Sydney, Australia, the lead author of this poster.

The findings were presented on 10 October at the ESMO 2016 Congress in Copenhagen, Denmark.

This large biomarker analysis using pooled data from the CheckMate 069 phase II trial and the CheckMate 066 and 067 phase III trials wherein 832 treatment naive patients with melanoma were randomised to nivolumab at 1 mg/kg plus ipilimumab at 3 mg/kg every 3 weeks (Q3W) for 4 cycles (n = 358) or nivolumab at 3 mg/kg Q2W, followed by nivolumab at 3 mg/kg Q2W (n = 474) until progression or unacceptable toxicity. Tumour tissue from primary or metastatic sites was obtained at screening prior to treatment and assessed for PD-L1 expression using a validated Dako immunohistochemistry assay.

In the combination nivolumab plus ipilimumab treatment cohort 92 (26%) of patients had PD-L1 expression ≥5% and 139 (29%) patients receiving sole nivolumab had PD-L1 expression ≥5%. The vast majority (80%) of tumour samples with positive PD-L1 staining had PD-L1 expression levels at or below 10%.  Expression did not appear to differ based on the site of biopsy collection, nor between primary and metastatic lesions. Using different thresholds, e.g. >1%, 5% and 10% of PD-L1 expression did not improve the ability of the test to discriminate between those patients most likely to gain a response to either treatment. In addition, using a receiver operator curve analysis did not define an optimal threshold that maximizes sensitivity and specificity for detecting responders. The area under the curve from this large number of patients treated with nivolumab was 0.6. A truly useless diagnostic test e.g. one that is no better than identifying true “positives” has an area of 0.5, whereas a “perfect” test would have an area of 1.0.

In the subgroup of patients with PD-L1 expression ≥5% median PFS was not reached (NR) in the combination treatment compared with 22.0 months for patients receiving sole nivolumab. However, with a minimum follow-up of 18 months, the hazard ratio (0.99; 95% Confidence interval [CI 0.66, 1.46) was similar between both treatments suggesting there was no difference in PFS benefit between both agents.

In the subgroup of patients with low (<5%) to no PD-L1 expression, median PFS was 11.1 months in patients receiving nivolumab plus ipilimumab compared with 4.9 months for nivolumab (HR 0.70; 95% CI 0.57, 0.87).

By contrast to the PFS results, the overall response rate (ORR) was higher in the combination treated patients, regardless of PD-L1 expression. In patients with PD-L1 expression ≥5%; the ORR in patients receiving combination was 68.5% compared with 59.0% in the nivolumab group. In the subgroup of patients with <5% PD-L1 expression, the ORR was also higher at 54.9% versus 39.7%. The median duration of response (DoR) with nivolumab plus ipilimumab was NR in both PD-L1 expression subgroups. In patients receiving only nivolumab, the DoR was NR versus 22.3 months in patients with ≥5% PD-L1 and <5% PD-L1 subgroups, respectively.

Conclusion

The subgroup of patients having high PD-L1 demonstrated similar median PFS with combination and nivolumab treatment alone. However, the ORR with nivolumab plus ipilimumab was numerically higher across both PD-L1 expression subgroups. The authors cautioned against using these results to assess the relative benefit of nivolumab plus ipilimumab versus nivolumab, as the OS data remains immature, and the use of PD-L1 as a biomarker remains problematic, and other biomarkers are emerging as potentially more important predictors of response. Specifically, PD-L1 testing is not a binary test, such as BRAF mutational testing, and several factors can contribute to false-positive results, such as the age of the sample, the use of different tests, and as a result of manual histological scoring. Dr. Long commented that decisions about different treatment options should be made on an individual patient basis, with clinicians accepting that full understaing of PD-L1 is far from complete.

Reference

1112PD

PD-L1 expression as a biomarker for nivolumab (NIVO) plus ipilimumab (IPI) and NIVO alone in advanced melanoma (MEL): A pooled analysis

G.V. Long, J. Larkin, P.A. Ascierto, F.S. Hodi, P. Rutkowski, V. Sileni, J. Hassel, C. Lebbe, A.C. Pavlick, J. Wagstaff, D. Schadendorf, R. Dummer, D. Hogg, J.B.A.G. Haanen, P. Corrie, C. Hoeller, C. Horak, J. Wolchok, C. Robert